Background
Acute myeloid leukemia (AML) is a highly malignant cancer of the bone marrow that is characterized by the rapid growth of abnormal myeloid cells. In 2012, leukemia accounted for 2.5% and 3.2% of all new cancer cases and deaths worldwide, respectively [
1]. The incidence rate of leukemia increased from 4.7 to 5 cases per 10,000 Koreans from 1999–2010 [
2]. AML cytogenetic studies provide important diagnostic and prognostic information for AML patients. However, approximately 50% of AML patients have a normal karyotype (NK-AML). Although Schlenck et al. showed that the combination of the mutations in
FLT and
NPM1 or CCAAT/enhancer binding protein (C/EBP), alpha (
CEBPA) could be used to predict NK-AML prognosis, most patients did not have this mutation set [
3]. This finding suggests that AML is a highly heterogeneous disease and that a large number of causal mutations have not yet been uncovered [
4]. Two systems, the French-American-British (FAB) classification and the newer World Health Organization (WHO) classification, have been used to classify AML into subtypes. To compare our results with the previous studies, we used the FAB system, the most commonly used classification in previous sequencing studies for AML, including The Cancer Genome Atlas (TCGA) whole genome sequencing (WGS) [
5,
6]. AML is classified into eight subtypes (M0 through M7) according to the FAB classification based on its morphological features: early forms in white blood cells (M0-M5), red blood cells (M6), and platelets (M7) [
3].
Next-generation sequencing technologies have extended AML genetic studies to a genome-wide scope at a single-base resolution. Ley et al. (2008) performed the first AML WGS study on one Caucasian woman with the cytogenetically normal AML subtype M1 and reported non-synonymous single nucleotide variants (nsSNVs) in eight genes (i.e.,
CDH24,
PCLKC,
GPR123,
EBI2,
PTPRT,
KNDC1,
SLC15A1, and
GRINL1B) and insertions in the coding regions of the
FLT3 and
NPM1 genes [
7]. Mardis and his colleagues reported that 16% of 80 NK-AML patients had a somatic point mutation in the
IDH1 gene [
8]. According to a later report by Ley
et al. (2010), approximately 22% of 281 AML patients had
DNMT3A mutations that were newly discovered using targeted sequencing [
9]. In a Chinese study using targeted exome sequencing, the patients with the
DNMT3A Arg882 mutation showed poor prognosis among AML-M5 individuals [
10]. A WGS study with eight Caucasian AML patients showed clonal evolution patterns and mutations associated with relapsed AML in the genes
WAC,
SMC3,
DIS3,
DDX41, and
DAXX [
11]. Recently, TCGA analyzed 50 and 150 patients with
de novo AML using WGS and whole-exome sequencing (WES), respectively. They found 23 significantly mutated genes and 237 gene mutations that recurred in at least two patients, which were grouped into nine categories according to their biological functions [
6].
AML is a clinically and genetically heterogeneous disease, hence discovering subtype-specific mutations may provide additional prognostic information for AML patients. In this study, we aimed to replicate previous findings in the European studies and to characterize the landscape of somatic mutations present in Korean acute myeloid leukemia. We also performed a stratified analysis for FAB M2- and M3-subtypes to investigate if certain mutations have subtype-specific effects. We subsequently evaluated the functional properties of the significantly mutated genes using an integrated systems analysis of Gene Ontology (GO) and biological pathways.
Discussion
A total of 11 genes were significantly mutated in 36 AML patients (
p < 0.01 and
q < 0.1). Specifically,
NEFH, hepatitis A virus cellular receptor 1 (
HAVCR1, 5q33.2), interferon, alpha-inducible protein 27 (
IFI27, 14q32),
PCSK5, and the arylsulfatase D gene (
ARSD, Xp22.3) have been implicated in a variety of cancers. For instance, the variants of
NEFH gene, a tumor suppressor, were suggested as prognostic markers for renal cell carcinoma (RCC) and contributed to susceptibility of esophageal squamous cell and hepatocellular carcinomas [
24,
36]. While overexpression of this gene interrupts the development of cell structure and function in normal cells [
37], loss-of-function mutations in this gene activate the Akt/β-catenin pathway and cause increased glycolysis and result in mitochondrial dysfunction in cancer cells [
24]. The transmembrane protease, serine 13 gene (
TMPRSS13, 11q23), a splice variant of mosaic serine protease large form (MSPL), encodes a family of the type II transmembrane serine proteases which plays critical roles in maintaining homeostasis, infection, and tumorigenesis [
38,
39]. The
HAVCR1 gene is a biomarker for diagnosing renal cell, ovarian, and colorectal carcinoma [
25,
40]. Especially, elevated expression of this gene prevents cancer cell invasion and adhesion in colorectal cancer cells [
25].
IFI27, the most highly up-regulated gene in human whole blood, is related with immune response through activation of T lymphocytes and dendritic cells [
41]. Furthermore, this gene induces the interferon-alpha and stimulates myeloid dendritic cells [
42]. The proprotein convertases (PCs) play important roles in development and metastasis of multiple cancers. The
PCSK5 gene (also known as PC5 or PC6) has been reported to be systematically down-regulated in intestinal tumors of the knockout mouse model and human [
43].
The protein encoded by the
ARSD gene located on X chromosome is a member of the sulfatase family, which is an essential element for skeletal and cartilage growth. The elevated expression of this protein was suggested to be associated with lipid metabolism such as sphingolipid and development of chronic lymphocytic leukemia [
30]. A cluster of Mastermind-like (MAML) genes, including
MAML1,
MAML2, and
MAML3, encodes transcriptional co-activators for various signal pathways such as Notch signaling and tumor suppressor pathway activated in multiple cancers [
44]. Specifically, the
MAML3 gene regulates the retinoic acid gene, which inhibits growth of human tumor cells [
45]. We identified four novel genes,
KRTAP4-5,
OR2T35,
GPRIN1, and
MRPL18, of unknown function in tumor progression and metastasis.
The genes mutated exclusively in the patients with the AML subtypes M2 and M3 have been reported more frequently in previous studies to have an association with AML and/or other types of cancers than the genes identified in the total AML patient group. For instance, previous NGS studies for AML evaluated the prognostic impact of the gene
CEBPA in AML patients [
6,
32]. This gene had the lowest
p value in the AML M2 group in the current study. The high proportion of M2-patients in the study subjects may have affected the results of previous studies. The E1A binding protein p400 (
EP400, 12q24.33) was reported to be associated with cancers such as RCC and colorectal cancer. All three genes specific to the M3 subtype (
ATXN3, thymine-DNA glycosylase (
TDG, 12q24.1), and
HCLS) also showed possible associations with cancer risks, such as B-cell chronic lymphocytic leukemia (B-CLL).
Gene set enrichment analysis showed that MAPK signaling pathway was significantly enriched in the 36 Korean AMLs. MAPKs play important role in converting extracellular stimuli into cellular responses and are often altered in cancers [
46]. The genomic region, Chromosome 11:117789342–117789345, harbor five missense mutations leading to protein changes, p.A77G and p.Q78R, of transmembrane serine proteases that is known to function in tumorigenesis [
39]. We additionally investigated the clinical features of patients who share the same recurrent mutations, including age, absolute neutrophil count, bone marrow blast percentage, white blood cell count, platelet count, and overall survival time, however there were no specific clinical features observed among them.
Acknowledgments
The authors thank all participants who donated samples and the staff members who were involved in this research. We especially acknowledge the contributions of Young Ki Lee (Syntekabio, Inc., Republic of Korea) who worked on NGS data generation, and Eun Pyo Hong (Hallym University College of Medicine, Republic of Korea) who contributed to the evaluation of the functional importance of the gene mutations.